Representation and efficiency in a production system for speech understanding

  • Authors:
  • Donald L. McCracken

  • Affiliations:
  • Computer Science Department, Carnegie-Mellon University, Pittsburgh, PA

  • Venue:
  • IJCAI'79 Proceedings of the 6th international joint conference on Artificial intelligence - Volume 1
  • Year:
  • 1979

Quantified Score

Hi-index 0.00

Visualization

Abstract

The research question studied is whether it is practical to use a production system (PS) architecture for a version of the Hearsay-II (IISII) speech understanding system. The study focuses on adequacy of representation, space efficiency and time efficiency A running PS architecture called HSP was designed and implemented and twelve HSH Knowledge sources were translated to HSP productions, with two of the twelve actually run under MSP. Detailed comparisons of HSH and HSP produced the following results: (1) HSP is adequate to represent the HSII speech Knowledge, even though HSP is a comparatively simple PS architecture; (2) HSP suffers a moderate space penalty for representing declarative HSII Knowledge, which is serious since HSII contains a high proportion of such Knowledge; (3) representation of HSII declarative Knowledge as HSP productions causes problems with multiple use of that Knowledge; (4) lack of a local worKing memory in the HSP architecture has serious consequences both for space and time efficiency; and (5) HSP has a time efficiency handicap of two to three-and-a-half orders of magnitude, in spite of efficiency mechanisms which make HSP comparable in time efficiency to other PSs. The paper ends with a set of questions that remain to be answered for a thorough evaluation of a PS version of HSII.